R Notebook

Our goal here is to look at the large and small particle size data.

Tempted to focus on station 043 again, so I can have fewer points and save my color coding.

library(tidyverse)
library(cowplot)
library(plotly)
pass <- function(x){x}

Load all data

unbinned_DepthSummary <- read_csv("dataOut/unbinned_DepthSummary.csv")
unbinned_EachSize <- read_csv("dataOut/unbinned_EachSize.csv")
binned_DepthSummary <- read_csv("dataOut/binned_DepthSummary.csv")
binned_EachSize <- read_csv("dataOut/binned_EachSize.csv")

Binned

binned_class_DepthSummary <- binned_DepthSummary %>%
  select(depth, profile, tot_TotParticles:big_speed, small_icp:big_psd) %>%
  pivot_longer(cols = c(-depth, -profile)) %>%
  separate(name, c("class", "variable"), extra = "merge") %>%
  pivot_wider(names_from = variable)

binned_class_DepthSummary %>% head

binndPlotable <- binned_class_DepthSummary %>% filter(profile == "stn_043", class != "tot")

plot_nparticles <- binndPlotable %>% ggplot(aes(x = nparticles, y = depth, shape = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 2500, by = 100)) + scale_x_log10() + guides(colour = FALSE) + labs(x = "#Particles/L") + theme_cowplot()

ggplotly(plot_nparticles)

# ggplotly(plot_biovolume)
# ggplotly(plot_flux)
# ggplotly(plot_speed)
# ggplotly(plot_psd)
# 
# cowplot::plot_grid(plot_nparticles,plot_biovolume, plot_flux, plot_speed, plot_psd)

long_binned <- binned_class_DepthSummary <- binned_DepthSummary %>%
  select(depth, profile, tot_TotParticles:big_speed, small_icp:big_psd) %>%
  pivot_longer(cols = c(-depth, -profile)) %>%
  separate(name, c("class", "variable"), extra = "merge") %>%
  filter(class != "tot", profile == "stn_043", !variable %in% c("nnparticles", "flux", "TotParticles"))

head(long_binned)

longPlt <- long_binned %>% ggplot(aes(x = value, y = depth, shape = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 2500, by = 200)) + scale_x_log10() + guides(colour = FALSE) + labs(x = "Value") + theme_cowplot() + facet_wrap(~variable, scales = "free_x", ncol = 2)
longPlt

We would have to plot these seperately to have appropriate x labels.

Fun fact. The bulk of flux is from small particles.

PSD

long_psd_binned <- binned_class_DepthSummary <- binned_DepthSummary %>%
  select(depth, profile, small_icp:big_psd) %>%
  pivot_longer(cols = c(-depth, -profile)) %>%
  separate(name, c("class", "variable"), extra = "merge") %>%
  filter(class != "tot", profile == "stn_043", !variable %in% c("nnparticles", "flux", "TotParticles"))

longPsdPlt <- long_psd_binned %>% ggplot(aes(x = value, y = depth, shape = class, col = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 2500, by = 200)) + scale_x_continuous() + guides(colour = FALSE) + labs(x = "Value") + theme_cowplot() + facet_wrap(~variable, scales = "free_x", ncol = 2)
longPsdPlt

Huh. There is a general flattening of the curve. That is, the large particles generally have a steeper PSD than the smaller particles.

long_psd_binned <- binned_DepthSummary %>%
  select(depth, profile, small_icp:big_psd, psd, icp) %>%
  pivot_longer(cols = c(-depth, -profile)) %>%
  separate(name, c("class", "variable"), extra = "merge", fill = "left") %>%
  mutate(class = replace_na(class, "tot")) %>%
  filter(profile == "stn_043") %>%
  #pivot_wider(names_from = variable) %>%
  pass

longPsdPlt <- long_psd_binned %>% ggplot(aes(x = value, y = depth, shape = class, fill = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 2500, by = 200)) + scale_x_continuous() + guides(colour = FALSE) + labs(x = "Value") + theme_cowplot() + facet_wrap(~variable, scales = "free_x", ncol = 2) + scale_shape_manual(values = 21:25) +
scale_fill_manual(values = c("tot" = "black", "big" = "blue", "small" = "red"))
longPsdPlt

• scale_fill_manual(values = c(“black”, “blue”, “red”), breaks = c(“tot”, “big”, “small”))

Something is screwing up my legend, but the fact is the small particles define the PSD. Big particles tend to have steeper PSD, indicating a flattening of the curve at smaller values.

https://stackoverflow.com/questions/21865207/adding-legend-ggplot-doesnt-work

Debug

test <- long_psd_binned %>% filter(depth < 200, variable == "psd") %>% select(-c(profile, variable))
test %>% head

https://stackoverflow.com/questions/43266636/incorrect-shape-and-fill-of-ggplot-legend

longPsdPlt <- test %>% ggplot(aes(x = value, y = depth, shape = class, fill = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 200, by = 50)) + scale_x_continuous()+ labs(x = "Value") + theme_cowplot()  + scale_shape_manual(values = 21:25) +
scale_fill_manual(values = c("black", "blue", "red"), breaks = c("tot", "big", "small"), labels = c("tot", "big", "small"))
longPsdPlt

longPsdPlt <- test %>% ggplot(aes(x = value, y = depth, shape = class, fill = class)) + geom_point(alpha = 0.5, size = 3) + scale_y_reverse(breaks = seq(from = 0, to = 200, by = 50)) + scale_x_continuous()+ labs(x = "Value") + theme_cowplot()  + scale_shape_manual(values = 21:25) +
scale_fill_manual(values = c("tot" = "black", "big" = "blue", "small" = "red"))
longPsdPlt